From fb86b97300d930b57471068720c52bfa8622eab7 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Tue, 18 Nov 2014 10:46:57 +0100 Subject: x86, microcode: Update BSPs microcode on resume In the situation when we apply early microcode but do *not* apply late microcode, we fail to update the BSP's microcode on resume because we haven't initialized the uci->mc microcode pointer. So, in order to alleviate that, we go and dig out the stashed microcode patch during early boot. It is basically the same thing that is done on the APs early during boot so do that too here. Tested-by: alex.schnaidt@gmail.com Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=88001 Cc: Henrique de Moraes Holschuh Cc: Fenghua Yu Cc: # v3.9 Signed-off-by: Borislav Petkov Link: http://lkml.kernel.org/r/20141118094657.GA6635@pd.tnic Signed-off-by: Thomas Gleixner diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c index dd9d619..2ce9051 100644 --- a/arch/x86/kernel/cpu/microcode/core.c +++ b/arch/x86/kernel/cpu/microcode/core.c @@ -465,6 +465,14 @@ static void mc_bp_resume(void) if (uci->valid && uci->mc) microcode_ops->apply_microcode(cpu); + else if (!uci->mc) + /* + * We might resume and not have applied late microcode but still + * have a newer patch stashed from the early loader. We don't + * have it in uci->mc so we have to load it the same way we're + * applying patches early on the APs. + */ + load_ucode_ap(); } static struct syscore_ops mc_syscore_ops = { -- cgit v0.10.1 From 02ecc41abcea4ff9291d548f6f846b29b354ddd2 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Sun, 30 Nov 2014 14:26:39 +0100 Subject: x86, microcode: Limit the microcode reloading to 64-bit for now First, there was this: https://bugzilla.kernel.org/show_bug.cgi?id=88001 The problem there was that microcode patches are not being reapplied after suspend-to-ram. It was important to reapply them, though, because of for example Haswell's TSX erratum which disabled TSX instructions with a microcode patch. A simple fix was fb86b97300d9 ("x86, microcode: Update BSPs microcode on resume") but, as it is often the case, simple fixes are too simple. This one causes 32-bit resume to fail: https://bugzilla.kernel.org/show_bug.cgi?id=88391 Properly fixing this would require more involved changes for which it is too late now, right before the merge window. Thus, limit this to 64-bit only temporarily. Signed-off-by: Borislav Petkov Link: http://lkml.kernel.org/r/1417353999-32236-1-git-send-email-bp@alien8.de Signed-off-by: Thomas Gleixner diff --git a/arch/x86/kernel/cpu/microcode/core.c b/arch/x86/kernel/cpu/microcode/core.c index 2ce9051..08fe6e8 100644 --- a/arch/x86/kernel/cpu/microcode/core.c +++ b/arch/x86/kernel/cpu/microcode/core.c @@ -465,6 +465,7 @@ static void mc_bp_resume(void) if (uci->valid && uci->mc) microcode_ops->apply_microcode(cpu); +#ifdef CONFIG_X86_64 else if (!uci->mc) /* * We might resume and not have applied late microcode but still @@ -473,6 +474,7 @@ static void mc_bp_resume(void) * applying patches early on the APs. */ load_ucode_ap(); +#endif } static struct syscore_ops mc_syscore_ops = { -- cgit v0.10.1